Center for Uncertainty Studies Blog

Tina Comes (TU Delft) showed examples from her research on the spread of the Covid virus in the Netherlands. © Adrian Strothotte / Bielefeld University.

From March 12–14, 2025, the conference "Building Models of Change: Bridging Sciences and Humanities" took place at the Center for Interdisciplinary Research (ZiF) in Bielefeld. The event brought together researchers from diverse disciplines to explore how different fields conceptualize and model change. A key focus of the conference was the role of uncertainty in these modeling efforts.

Section 3, titled Uncertainty in Modeling and curated by the CeUS Founding Directors Silke Schwandt and Herbert Dawid, provided a dedicated space to examine how different disciplines engage with uncertainty. Presentations explored uncertainty in historical knowledge production, clinical psychology, economic decision-making, and resilience modeling. The discussions revealed both the challenges and opportunities of incorporating uncertainty into models across various domains.

Historiographical Uncertainty and Computational Historical Models

In his presentation, Prof. Dr. Michael Piotrowski (University of Lausanne) emphasized that all human knowledge is uncertain, inexact, and partial (Russell, 1948). His talk focused not on historical uncertainty per se, but on historiographical uncertainty—the challenges historians face in reconstructing the past. Piotrowski compared the approach of the historian to that of a detective, constructing causal models and narratives based on fragmented evidence. Piotrowski noted that even if historians had access to all possible sources, uncertainty would persist due to the interpretative nature of historical writing. Digital humanities methods, including statistical models, digital reconstructions, and TEI-based critical editions, offer new tools to address these uncertainties, yet they cannot eliminate them. He underscored the limitations of computational modeling in the humanities, arguing that while digital humanities can employ models, their application is not as straightforward as in the natural sciences.

The discussion touched on the epistemological implications of modeling in history. Participants debated how models can represent causal sequences, the role of corpora as models, and whether historians should attempt counterfactual modeling. The conversation also explored the relationship between uncertainty and digital tools, questioning how different types of uncertainty—such as selection bias in sources—should be accounted for in "computational history" as Piotrowski liked to call it.

Individual Responses to Uncertainty and Broader Change Processes

Moving from historiography to psychology, Prof. Dr. Mark Freeston (Newcastle University) examined how insights from clinical psychology can inform broader understandings of uncertainty. His work on Intolerance of Uncertainty (IU) has demonstrated how people’s responses to uncertainty shape their cognitive and emotional processes. Originally developed in the context of generalized anxiety disorder (GAD), Freeston expanded the concept to explore decision-making, risk perception, and even political behavior. Freeston emphasized that uncertainty is often the absence of safety, a distinction that has implications beyond clinical settings. His studies suggest that fostering tolerance of uncertainty can enhance resilience and adaptability. He illustrated how structured interventions—such as transdiagnostic group therapy—can help individuals engage more constructively with uncertainty. A key takeaway from his talk was the societal relevance of these insights. Uncertainty is a driving force in public discourse, from pandemic responses to political polarization. Freeston argued that owning uncertainty—rather than attempting to eliminate it—can build trust and improve decision-making.

This perspective sparked a lively discussion on the communication of uncertainty in science, politics, and education. Participants also reflected on cultural differences in how uncertainty is perceived and addressed, noting that e.g. the Dutch and German languages lack distinct words for the difference of uncertainty and insecurity.

Modeling Decision-Making Under Uncertainty

Prof. Dr. Jean-Marc Tallon (Paris School of Economics) provided an economist’s perspective on uncertainty, presenting models of decision-making under uncertainty. His talk built on classic decision theories, such as Bayesian probability models and game theory, to examine how individuals and institutions make choices under uncertainty. A central question was whether all uncertainties should be treated probabilistically or if alternative approaches are needed. Tallon discussed smooth ambiguity preferences, a model that accounts for uncertainty aversion, and explored its applications in economics—ranging from financial markets to climate policy. He also addressed the challenge of distinguishing between uncertainty and risk, noting that traditional economic models often assume that uncertainty will diminish as more information becomes available. However, real-world scenarios frequently involve persistent uncertainty that resists quantification.

During the discussion, participants debated the applicability of economic models to social and historical phenomena. Questions arose about how well models derived from laboratory settings translate to real-world environments and whether alternative modeling strategies—such as reinforcement learning—could better capture decision-making dynamics.

Temporal Dynamics of Resilience Modeling

The final talk in Section 3 (and of the conference in general) was given by Prof. Dr. Tina Comes (TU Delft), who examined how models can support decision-making in crisis situations. She emphasized that uncertainty is not merely a challenge to be mitigated but a fundamental feature of complex systems. Her research on resilience modeling integrates data-driven and expert-based approaches to navigate uncertainty in disaster response, urban planning, and climate adaptation. Comes highlighted the difficulties of balancing short-term urgency with long-term strategic planning. Using examples from crisis management in the Philippines, climate resilience in the Netherlands, and urban development in Helsinki, she demonstrated how past decisions shape present vulnerabilities. She warned that models must be used carefully, as over-reliance on predictive tools can create a false sense of certainty.

The discussion explored the role of historical data in resilience modeling, the feasibility of cross-cultural comparisons, and strategies for evaluating models over long time horizons. Participants also considered the ethical implications of decision-support models, particularly in contexts where uncertainty intersects with political and economic power structures.

Modeling Uncertainty Across Disciplines

The presentations in Section 3 illustrated the diverse ways in which uncertainty is conceptualized and modeled across disciplines. Across all fields, participants acknowledged that uncertainty is not just a technical problem but an epistemological and practical one. Effective modeling requires an awareness of the assumptions embedded in models, the limits of prediction, and the broader social contexts in which uncertainty operates. By bringing together scholars from different disciplines, the conference facilitated a productive exchange on how uncertainty can be studied, communicated, and incorporated into models. The discussions underscored the need for continued interdisciplinary dialogue to refine our approaches to uncertainty.

For more insights and upcoming events, stay tuned to the CeUS Blog.